Scalding Commons

(This page contains the README for the former scalding-commons library. All scalding-commons code has been merged into the main Scalding repo as of June 6th, 2013.)

Common extensions to the Scalding MapReduce DSL.

Dfs-Datastores Integration

Scalding-Commons includes Scalding Sources for use with the dfs-datastores project.

This library provides a VersionedKeyValSource that allows Scalding to write out key-value pairs of any type into a binary sequencefile. Serialization is handled with the bijection-core library’s Injection trait.

VersionedKeyValSource allows multiple writes to the same path,as write creates a new version. Optionally, given a Monoid on the value type, VersionedKeyValSource allows for versioned incremental updates of a key-value database.

import com.twitter.scalding.source.VersionedKeyValSource
import VersionedKeyValSource._

// ## Sink Example

// The bijection library provides implicit Injections
// from String -> Array[Byte] and Int -> Array[Byte].
val versionedSource = VersionedKeyValSource[String,Int]("path")

// creates a new version on each write

// because Scalding provides an implicit Monoid[Int],
// the writeIncremental method will add new integers into
// each value on every write:

// ## Source Examples
// This Source produces the most recent set of kv pairs from the VersionedStore
// located at "path":

// This source produces version 12345:
VersionedKeyValSource[String,Int]("path", Some(12345))